“We will never shut down the country again. Never.” President Trump’s tone was emphatic, edged with agitation. Furrowing his brow, he concentrated his full attention on me. His pupils narrowed into hardened points of anger.
We were standing in the narrow five-by-eight-foot space just outside the formal White House Briefing Room, which was crammed with hardworking communications staff. It was the first week of April 3, mere days after the president had announced the thirty-day extension of the Slow the Spread campaign to the American public, and the ground had shifted suddenly and without warning.
I felt the blood drain from my face, and I shivered slightly.
A moment later, I stepped into another press briefing, swept along by the frigid wake coming from the president’s broad back. I experienced that unnerving sensation of having crested a hill too quickly while driving.
As tempting as it was to believe that this was a belated April Fool’s joke, I knew it wasn’t. I didn’t know what precisely had brought about his change of heart, or who had convinced him I was wrong, but his belief in me—in the science, the analyses, the graphs that had gotten the thirty-day extension approved—seemed to have disappeared nearly overnight. His stern look suggested I’d betrayed him, misled him somehow. I didn’t take his anger personally—three years of watching him on television had trained me not to be dumbfounded by his words or actions. Still, the whiplash was intense.
After a month of positive and relentless forward progress, President Trump’s harsh words and ominous tone immediately cast a dark shadow over the country’s future. What interests had driven this new push? What behind-the-scenes actors had influenced him? More concerning: What were we going to do to prevent the calamity I’d predicted from coming to pass?
I assessed the situation: Okay, we’ve gotten forty-five days. We’re not going to get any more. How else can we protect people moving forward? What other ways can we push this rock back up the hill?
We were less than a week into our second circuit breaker, protecting the metros where the virus wasn’t in full community spread. While I’d always assumed getting another shutdown would be a near impossibility, having the door slammed shut so hard, so fast would leave us with few options if the situation on the ground continued to deteriorate. If we ended the shutdown too early and the American public’s behaviors reverted to pre-pandemic devil-may-care, thousands more could die. These additional thirty days would be critical, and slow, careful reopening down the line would be essential.
What I couldn’t have known then was that this day would mark a permanent change in my relationship with President Trump. His about-face created a seismic shift in my ability to speak directly to, present data directly to, and influence him in person. Though, in theory, data-based decision making was still possible in this White House, from here on out, everything I worked toward would be harder—in some cases, impossible.
With the president’s harsh words still ringing in my ears, I walked into the April 7 press conference and took my place on the dais, standing off to one side. I wondered if the makeup I’d been wearing to hide my exhaustion was adequately masking the “What-the-fuck-just-happened?” expression on my face. I had come into the White House knowing that having the president’s ear would be crucial to my success. I’d never suspected how suddenly I could lose it, how impossible it would be to get it back, nor how sweeping the impact of his ghosting me would be for Americans.
With the cameras trained on all of us, I maintained my composure.
And with that, April began.
IN APRIL 2020, NEARLY everything came undone. So much of what went on that month codified the president’s instincts about the virus and about the “cure” being worse than the disease. The ongoing shutdown was good news: we had preserved whatever positive strides the states had made in March. But behind the scenes, the administration was laying groundwork for a radical change. It didn’t matter what I or any of the scientists thought: whether cases were growing by 10 percent or 300 percent, whether testing had improved, whether more people were dying. From the start of April, so long as Donald Trump remained the chief executive, the federal response would be different, and I would have to adapt to effectively protect the country from the virus that had already silently invaded it.
Since my arrival, the demarcation line between the economic and public health interests had been clearly defined. I always knew that President Trump was influenced by whomever he’d last spoken with. What I couldn’t yet decipher was whether his new attitude was coming from the political side (Mark Meadows, Marc Short, Peter Navarro, Derek Kan, Derek Lyons, and Stephen Miller) or the economic side (Tyler Goodspeed and Kevin Hassett). To some extent, the question was one of semantics: with this White House, all politics was about the economy. Still, the distinction mattered. If I was going to watch my back, I would have to identify who the biggest threats countering my analyses were.
Tuesday, March 31, was Mark Meadows’s first day in the office as the president’s chief of staff. I hadn’t actually met Meadows. He’d attended the 2020 Conservative Political Action Conference (CPAC) and had potentially been exposed to someone with Covid-19. Just before I went to a Sunday task force meeting on March 8, he called me from home quarantine. He was clearly unhappy about the CDC’s recommendations. He saw little scientific evidence for the length of quarantine or for the incubation period of the virus the CDC had cited. The data set was too small at that point, he said, for such strict measures. To be clear: the data set was extremely limited. It was based on a choir in Washington State and didn’t take into account that the members infected on day ten and fourteen may have been so from another source. Sequencing of the virus from each choir member hadn’t been done, so it was possible those late cases were unrelated to the choir’s exposure.
First impressions are revealing, and right away, I put Meadows in the “the cure is worse than the disease” camp—but with a twist. Unlike Marc Short, who too easily dismissed any claim the scientists made, Meadows wanted to see the full data set and hear the rationale behind our interpretations. He was willing to have an open debate. In later conversations with him, he often asked me for the raw data used for our recommendations. I was always suspicious about whom he shared these figures with, what additional analyses were done, and how they were used. Still, he was the president’s chief of staff; if he requested something of me, I assembled the data. At the time, sharing numbers was part of my job. I couldn’t yet see how it would also be a part of my undoing.
On April 3, just minutes removed from President Trump’s emphatic declaration, it was impossible to dissect his message with great assurance, but the trail of breadcrumbs seemed to start with the projections I had produced to get the additional thirty days: the 100,000–240,000 deaths. I suspected that as far back as the European travel ban debate, when I asked the economy people where their numbers were, someone within the administration had been reinterpreting the data sets I had used for my projections. Their goal was simple but dangerous: to let the president believe I was wrong in my calculations and therefore couldn’t be trusted.
On April 8, I received a memo on which I was cc’d. It came from the CEA’s Tyler Goodspeed—whose earlier memo, you’ll remember, had stated that any interventions we made against the viral spread would be only 20 percent effective. Kevin Hassett had asked Tyler to turn his attention again to the response to the pandemic. Hassett had once served as the president’s CEA chairman. He’d left that position but was then brought back to President Trump’s White House when the pandemic response began. Just like Secretary Mnuchin, both Goodspeed and Hassett had demonstrated a brilliant understanding of economics. Collectively, they and others at the CEA had built trust and respect within the West Wing. They had credibility—something I’d apparently earned and then quickly lost. Yet, neither had dealt with a vast public health crisis of this nature before. For all Kevin Hassett’s excellent ability to integrate economic data, he and Tyler were in new and complicated territory.
With Tyler’s April 8 memo, I had my answer for who within the senior administration was at work countering my data. Kevin Hassett, Tyler Goodspeed, and other members of the CEA team had put together an independent analysis of the data I’d used for my computational assumptions, using similar case curves they’d developed. The problem was that they used vastly different assumptions than the ones I had used. Italy and many European countries had reached a peak for new infections, Tyler wrote. “We then adjust these projections to U.S. population level and apply country-specific Case Fatality Rates (CFR) to cumulative population adjusted projected cases. Specifically, extending each country’s curve using a cubic model and scaling it to the U.S. population.” He continued:
Approximately 104,000 U.S. deaths, assuming the highest observed CFR (Italy, 12.9 percent), with a high estimate of 147,000 deaths in the event the U.S. were to track Spain in cases and Italy in CFR. Assuming the actual CFR observed in the U.S. to-date suggest substantially fewer cumulative deaths (approximately 26,000).
Their conclusion: in this first surge, 26,000 people would die by Memorial Day. We had predicted between 100,000 and 240,000. Clearly, this was an enormous difference. My projections were nearly four to ten times greater. The CEA and I saw two very different futures. I saw a pandemic of historic proportions; they saw a fairly average year of seasonal flu.
On the day I received the memo, April 8, the United States had already had 18,000 fatalities. Did the CEA really believe at this stage, given how the numbers were trending, that only 8,000 more people would die in eight weeks? By April 12, literally four days after the memo was written, the country surpassed the 26,000 deaths Tyler and his team had predicted. On April 8 alone, 2,234 people perished from Covid-19; the spring surge topped out on April 21 at 2,725 fatalities per day and stayed above 2,000 for seventeen more days. Obviously, this was still a long way from Memorial Day.
The underestimation and underselling of the seriousness of the outbreak’s progression was obvious to me. They were trying to meet data with data. But they weren’t using the appropriate data assumptions to arrive at their figure. Not obvious was who was responsible. I did what I usually do and checked Tyler Goodspeed’s credentials. He had published widely on banking and financial regulation. While the memo had come from him, I knew from meetings and conversations that the driving force behind it was Kevin Hassett, the expert economic forecaster. To be clear this work was not from Larry Kudlow or Secretary Mnuchin. They never contradicted my numbers or projectinos and instead used them.
They had used the same data I’d assembled, and they’d done the math right, but they’d used very different assumptions to come to conclusions that were very different from mine—and much more palatable to the Trump White House.
Their model wasn’t accurate for these reasons. First, the CEA presumed incorrectly that viral outbreaks in the major metropolitan areas would occur simultaneously and not serially. Second, they treated the demography of each major metropolitan area identically. And finally, and most crucially, they had grossly underestimated the United States case fatality rate, failing to take into account both the delay between infections and fatalities and between fatalities and their reporting.
Their projection was misinformed. I had used the Italian data because I was constantly in contact with that country’s public health people. From those discussions, it became apparent that Italy had more comprehensive and up-to-date data. Their figures included the most essential demographic information (age and comorbidities), and I was able to use their fuller data to project that our outcomes would closely match theirs. Our data-reporting system was always going to skew projections primarily due to lateness of reporting, and I was able to account for that. Because I knew our death data was delayed by weeks, the fatalities from today’s cases would happen and then be reported weeks in the future. Today’s reported deaths were from weeks prior when cases would have been exponentially lower.
The economic team’s faulty assumptions would always produce an underestimate. By using the wrong CFR, they would continually compound this fundamental miscalculation, day after day, case after case. They sent their report out with those wrong assumptions. They never bothered to consult with me. That was frustrating enough, but believing that a total projection of 26,000 dead when 18,000 had already perished was mind-boggling.
They had failed to understand a fundamental point I had tried to get across to them repeatedly: I had used the data from Italy to make clear that we would experience something like what Italy had experienced—but crucially, the United States wasn’t mitigating the spread as aggressively as Italy was, so our fatalities would assuredly be higher. Also, Americans suffer from comorbidities at a higher percentage than Italians, again contributing to a higher CFR. When you factored in these two variables (lower mitigation measures and poorer population health), the U.S. case fatality rate would absolutely be worse than Italy’s.
At first, I wanted to immediately dismiss the CEA’s forecast due to those fundamental errors. The flaws were obvious to me, but it would be harder for people without an epidemiological background to see them; they didn’t know that Italy reported deaths regularly, and we didn’t. You have to know your data sources. You have to know their limitations.
I suspected that the CEA’s faulty analysis and numbers had gone to the president and other senior White House advisors, like Jared Kushner, Marc Short, and Mark Meadows. The CEA had done its analysis and filed its report just days after I presented mine to the president and convinced him to authorize the thirty-day extension. Just as I’d been preparing daily reports, models, and projections, so had the CEA.
As gatekeepers for their respective bosses, Short and Meadows had control over who and what went to the vice president and president. The Hassett/Goodspeed analysis must have made its way immediately to the Oval Office. This president wasn’t going to care about the subtleties of the analysis. The more cynical of the White House advisors would have bottom-lined it for him. They’d likely reduced it to this:
The doctors say a hundred thousand to two hundred forty thousand dead.
Your economists say twenty-six thousand.
Debbi Birx used the wrong case fatality rates—she used Italy’s. We used the right one, the United States’.
You know we are better than Italy. Our hospitals are better.
You know she isn’t very good. She’s just another civil servant in over her head.
Reading between the lines, then:
Debbi Birx is wrong, really wrong.
She has overestimated, by four to ten times, the number who will die.
Debbi Birx intentionally misled you to get you to do something that was never needed, shutting down the country for another thirty days.
The number will be under thirty thousand—almost eerily close to the seasonal flu.
This will be an acceptable level of loss to the American people.
No need to worry. No need to treat the pandemic aggressively.
Trust us, not her.
We know better. We are your team. Together, we built your economy.
We know numbers. Our numbers are right.
While I had been open about my forecasts being based on the best available data, they had presented theirs under the guise of its being much more consistent with the real numbers, the American numbers. In fact, their flawed methodology did not, and was never going to, reflect reality.
It helped the CEA enormously that their data happened to jibe with a typical seasonal flu death count, magically matching what the president and many in the administration had been saying all along. As time went on, and especially in retrospect, I saw that as soon as the Hassett/Goodspeed report was circulated among the Oval Office leaders, the president had all he needed to confirm his initial bias that the novel coronavirus was just like the flu. With this confirmation in hand, President Trump, I believe, simply stopped focusing time and attention on the public health side of the pandemic response. Mistakenly, he no longer believed he needed to support any of the mitigation efforts that were key to slowing the spread. He’d moved on. A set of flawed numbers from his “best” people supported the notion that he’d been right all along.
In the moment, the importance of the 26,000 versus 100,000–240,000 discrepancy didn’t fully register with me. Over time, as the deaths continued to rise, the CEA would up its estimate, eventually getting to 66,000. That was still well below my projections but near enough to mine that we could have engaged in dialogue. I could have shown which of their errors explained the difference, how they had employed a similar model but with wrong assumptions. But no one initiated that conversation.
In April, my attention was being drawn to so many other areas that, regrettably, I didn’t try to reclaim the president’s attention. And I suppose it was too much to expect, given the atmosphere in this White House, for his most senior advisors, to say, “Our original estimate was wrong. Dr. Birx did a better job, and her figures are more representative of reality.” Admitting to an error is the right thing to do, but not in this administration.
I don’t think the CEA or Mark Meadows ever returned to this issue, and the flawed numbers were left uncorrected. I still wonder: What if I had demanded an audience with the president and Hassett? What if we had sat down and gone through the differing projections, side by side? Would doing so have changed the president’s mind? Would he have understood that I really did know what I was doing and could be trusted for objective facts and figures? I might have been able to alleviate the administration’s worst fears by suggesting that the optimal way forward wouldn’t necessarily “shut down” the country and that we could, in fact, maintain significant economic activity while still protecting Americans.
I don’t know. Today, I regret not having tried.
This admission doesn’t mean I did nothing to address the grievous differences in our respective projections. During the travel ban deliberations and the discussions over shutting down for fifteen or thirty days, the two wings of the task force (economists and medical professionals) had been urged to present our cases and arrive at consensus. Once I received the CEA report, I wrote to Tyler Goodspeed and Derek Kan. I pointed out that their math was good, but that we needed to discuss their underlying assumptions. I fully expected that we’d get together and that each side would present its case. This never happened.
When advisors cherry-pick data, without fully understanding the data or its sources, to paint a contrary picture of a pandemic to the president of the United States, it is not only intellectually dishonest, but also morally negligent. When the CEA didn’t respond to a request to work out our differences, they sent a powerful signal: We don’t care about your numbers, we care about supporting the president’s wishful thinking. With that single memo, they created the first of many inflection points in the president’s engagement with the pandemic response.
I saw this ongoing data discrepancy as a harbinger of future instances in which others would either step forward openly or work behind the scenes to undercut the scientists’ influence on the president and on policy making. At this stage, none of these efforts were overt. Mostly, they manifested themselves in the president’s no longer engaging with me or the other scientists on the task force. This was not the case for the vice president and the heads of the other agencies: They, and other members of the task force, stood together. They remained grounded in using data for decision making and, critically, in supplying public health information, equipment, and therapeutics to save as many lives as possible.
I didn’t think then, and I don’t believe now, that Steve Mnuchin or Larry Kudlow instigated the Hassett/Goodspeed report and the CEA’s inadequate projections and incorrect assumptions. During the debates on the European travel ban, and for a short while after the debate over the duration of the shutdown, they grilled me hard about our projections. They seemed generally convinced, if not by the numbers themselves then by the level of rigor we had applied to produce them. (They hoped the economic damage could be limited—as we all did.) After that, my interactions with Steve Mnuchin were limited to his participation in task force meetings, because he was off working on the legislative agenda to support the American people economically through the pandemic. I believed then and I believe now that Steve took the pandemic very seriously and understood the risk of Covid-19 disease to Americans. He also understood the economic impact it would have on people, and he worked 24/7 on a series of bills, and on the penultimate CARES Act, to get funding efforts through Congress.
When he was about to make press appearances, primarily on the Sunday shows, Larry Kudlow always came to me and said, “I’m going out there. What can you tell me about what’s going on?” I believe he presented a balanced approach. He would talk about what the administration was doing to respond to the economic situation through its policies and legislative agenda. He would also talk about the importance of the personal behaviors and measures needed to stem the community spread.
To further counterattack the effects of the economic advisor’s inaccurate projection, I sent Irum to work with the CEA folks on their daily report. No matter what she did, no matter how she advised them to address the three main problems with their analysis, she met resistance. Just as they believed that the president wouldn’t listen to us, they weren’t going to listen to her. They’d spoken. They’d supported the president’s view, and that was now that. He’d heard “twenty-six thousand,” and there was no way to make him un-hear it.
It wasn’t surprising that the CEA and those who believed their projections avoided having an actual discussion on the numbers—especially once my daily reports and summations of the data clearly showed that we surpassed their 26,000 deaths a mere four days after I received the Goodspeed memo. Tragically, by May 25, Memorial Day, nearly 100,000 Americans had lost their lives to Covid-19. Our projections were accurate; the CEA’s were not. But who among them was going to step up, in that environment, and admit they were wrong?
In late July, when Kevin Hassett returned from working on another assignment, he asked Tyler Goodspeed to again engage in this area in which he didn’t have any great expertise: modeling public health projections. Tyler did the right thing: he declined. I believe he’d seen how far short their figures had fallen and what had resulted from the CEA’s faulty report. After recognizing the consequences of his first attempts at pandemic modeling rather than economic modeling and predictions, Tyler didn’t want to repeat the same mistake—unusually humble behavior in this White House.
I sensed at the time of President Trump’s “never again” remark, and even more strongly with the CEA memo, that the brief window of opportunity I’d used to make my case for shutdown—the president’s fear for his own health when friends and contemporaries were on ventilators or dying—had now closed. A unique set of circumstances, a moment of vulnerability, had nudged me closer to the front and gotten the United States thirty more days of mitigation.
Somehow, from the time he agreed to the thirty extra days, the president had convinced himself both that he was physically invulnerable and that if he didn’t do everything he could to get the country (that is, the economy) back up and running at full speed, he would be politically vulnerable come November. The general election was just around the corner, and a robust economy was his ticket to four more years. Once the president was again numb to the devastating effects of the disease he’d seen ravage his friends, he was off and running in the other direction, leaving me in his dust. Even after falling ill himself, he would never again return to that Yellow Oval “Okay, okay” moment I’d witnessed.
I suspected then, and am now absolutely convinced, that by the time I got the CEA memo with its flawed assumptions, I had already been many days behind in the race for the president’s attention and trust. I am sure that in every internal senior advisor discussion, the president was reminded of just how wrong I had been in my projections. It never mattered that my projections were right then and continued to be right throughout 2020 and 2021; being right apparently didn’t matter. I know I find little consolation in it.
After my initial success with slowing the spread, I often felt I was just one chart, one statistic, one direct meeting with the president away from getting him back on my side. I believed that I was always this close to getting him to use clear data to drive mitigation efforts. If I had, perhaps he would have been willing to more forcefully advocate for basic mitigation efforts like masking and reducing the numbers of people dining indoors. He might have told the American people that gatherings of friends and family were one of the main causes of infection and that frequent and strategic testing could prevent a worsening of community spread. Reality, of course, was much crueler. We would have slowed the spread of the virus and significantly reduced the number of people who perished before Labor Day. Not being able to demonstrate to him how close the two sides, economic and medical, actually were in our numbers was, and remains, heartbreaking.
AS HARMFUL AS IT was to the public health response, the CEA team was just one front in the White House war on the medical professionals on the task force. As April would reveal, the marshaling anti-shutdown forces looked for any opportunity to undermine the data and metrics we were using to justify the seriousness of the situation—often pitting science against science.
Around that time, Jared Kushner shared an email with me. Up to that point, Jared had been mainly a peripheral figure, moving at the edges of my vision as he managed some of the essential logistics and supply issues related to PPE. He had understood the need to go directly to the suppliers and began what became the air bridge that brought essential gloves, gowns, and surgical masks directly from China, Malaysia, Thailand, and Vietnam throughout the late spring. To my mind, he was an effective means of getting to the president. I had no direct interaction with him on which to base any other assessment.
The memo Jared shared with me cited a National Institutes of Health analysis of the Italian data I’d used as a baseline for my projections for the case fatality rate. The memo stated that the NIH had determined that only 12 percent of the Covid-19-related deaths were directly attributable to the virus. Twelve percent? This was a ridiculous number. Now it seemed someone was pitting a trusted scientific organization, the NIH, against the task force coordinator. More significantly, they were hoping to erode the bedrock on which our case for shutdown had been built: the very reliable, precise, and timely information the Italians had provided. I immediately called Tony. He hadn’t seen the analysis and certainly didn’t support its conclusion.
I had had no idea the NIH was doing this study. If not for Jared’s heads-up, I could have been blindsided. Like Joe Grogan, he was warning me. In this instance and others, Jared Kushner let me know what was going on. Indirectly, through him, I was able to gain insight into what was taking place outside the task force, in the hallways and behind the doors of the president’s private dining room.
As he did throughout my White House tenure, Jared listened to the data I presented in support of my recommendations. I believed then, and I still believe, that he and the vice president advocated for my analyses and recommendations with the president—whether it was the need to expand testing, the importance of masking, or the critical message about viral spread among friends and family.
In a White House infamous for loud voices and backdoor meetings and where a general lack of discipline ruled the day, it was difficult to remain objective. When faced with so much randomness and uncertainty, could anything anyone said be trusted? And whom was the president trusting? The importance of what you had to say didn’t seem to matter. It was about access—access to the president. It was about who could get to the president and who had the last word.
It was also about where people were getting their information. When I reported fatality figures, I integrated the data from multiple reliable sources in the field—from health ministers across Europe; from the myriad Covid-19 data-tracking sites, like Our World in Data and Worldometer; from Johns Hopkins; and from hospitals and nursing homes. Stephen Miller, the White House director of speechwriting and one of the president’s most influential policy aides; Stephen’s wife, Katie Miller, Vice President Pence’s communications director; and Devin O’Malley, the vice president’s press secretary, all repeatedly claimed that the figures I’d provided were wrong: they were miscounts, they said; unverified accounting, they insisted. The Executive Office didn’t have its own source of data, so whom were they relying on to make this claim? What were they basing these claims on? They never presented me with evidence to support their positions.
Well, for one thing, there was no shortage of baseless claims pinging around the internet, and likely these fantastical stories were at least partially responsible for the “data” points the Millers and O’Malley had used to support their position that there were fewer Covid-19 cases and fatalities. Conservative radio host Wendy Bell, streaming live on Facebook, told her listeners that due to a change in death certificate procedures, “there’s a huge chance that Covid death numbers are exaggerated, to the tune of 94 percent.” This false claim mutated from there.
An immigration hard-liner, Stephen Miller used my insights and data when they overlapped with his focus on stricter border control. When I pointed out the increase in cases in Imperial, California, and El Paso, Texas, and other cities on both sides of the border, for example, Miller wanted to use this data to restrict border access. He believed that so-called illegals were responsible for the increase in cases. The CDC did a deep dive with city and county officials and the hospitals along the border where increased cases and hospitalizations occurred—and determined that American citizens and those with dual nationality residing in Mexico seeking care in the United States were driving up the numbers. But this bit of truth didn’t seem to matter. Miller and others used it as a wedge to further the divisiveness in the country over immigration and distract the public from the real problem at hand.
Mark Meadows also frequently challenged the hospitalization and fatalities figures I provided and questioned my sources. He listened to me, and we engaged in dialogue. At one point, he stated that many of the hospitalizations that had been coded as “Covid-19” admissions were actually the result of a SARS-CoV-2 test being administered after the patient had already been admitted because of an auto accident or for some other health reason. A Covid-19 diagnosis was an incidental finding. As such, those “after the fact” admissions shouldn’t have been counted toward the total number of Covid-19 cases.
Others suggested that hospitals added a Covid-19 code to their billing only to get the increased funding allocated to Covid-19 inpatient treatment. Later in my visits with hospital administrators, I learned that no hospital wanted more Covid-19 patients to care for and overwhelm their beds and ICUs. When this happened, the hospitals were forced to shut down elective procedures—the true moneymakers for them. Caring for Covid-19 patients is a complex, highly nurse- and physician-intensive struggle that uses up far more human resources than could ever be fully reimbursed. Even after I brought this information back to the task force, correcting their errors, social media postings continued to be seen as equally legitimate, factual sources as my on-the-ground on-site, truth-based findings.
On their own, these cases of incorrect and faulty data could easily have been dismissed as being endemic to our age and the rise of the internet as a source for all kinds of information, good, bad, and indifferent. But these claims were coming from people in the president’s inner circle, people influencing decision making at the highest level, decision making that would determine whether Americans lived or died. It was deeply troubling.
At about the same time Jared’s and others’ emails were crossing my desk, another influential group of researchers chimed in with results that further eroded confidence in my projections. Researchers affiliated with Stanford University and the University of Southern California had conducted a study using antibody tests to assess how many people diagnosed with Covid-19 disease based on symptoms had actually been infected with the SARS-CoV-2 virus.
When you are exposed to a virus, your body produces antibodies to fight it. An antibody test (as opposed to a diagnostic, swab-up-the-nose test, like the ones we’ve become accustomed to, which detect active Covid-19 infection) determines whether your immune system has produced antibodies specific to that virus. If the test is positive, you can be said to have been infected by the virus, whether you are asymptomatic or very mildly symptomatic.
In the study, researchers sought out volunteers in one county near San Francisco and others living in Los Angeles County. How those volunteers were selected and their reasons for participating could have created sampling bias, particularly if those included already suspected they had been infected and were looking for confirmation. The results in both counties “showed,” through the presence of SARS-CoV-2 antibodies in the blood, that far more people were infected with the virus than the diagnostic testing and data gathering had shown.
This sounds, on the surface, like a good thing for my case that silent spread was highly prevalent—and it was. News outlets picked up that portion of the story. But what the researchers had determined also undermined the conclusions we had drawn about the number of deaths that were likely to occur. One author of the study, in an interview with the New York Times, discussed their methodology, the results, and the conclusions drawn from them, and said, “This is very consistent with the fact that the virus is very common but not killing at the rate we thought.” Gina Kolata, a well-respected science reporter for the Times, wrote, “But the new data suggests most adults will experience milder to asymptomatic infections.” She went on to draw a reasonable conclusion that, if the study was accurate, then the fatality rate was more in line with the seasonal flu “than [with] a pandemic of profound lethality.”
Despite the study’s confirming that more people were infected with SARS-CoV-2 than previously believed, many in the White House chose to focus only on the much lower death rates and the flu fatality comparison and not on the silent spread that drove community spread and thus led to those deaths. They looked at this data and concluded: Covid-19 is not a big deal and won’t be a big deal. It is no worse than the seasonal flu, and we don’t lock down the country for the flu. They added this highly problematic study to the pile of evidence that Debbi Birx and, by association, Tony Fauci are wrong in their projections and views.
When the Stanford study was published in The Lancet, a highly regarded medical journal, this was another nail in the coffin of our projections and status. Stanford is a highly respected institution; so is Harvard Medical School, which publishes The Lancet. The message appeared to be: Trust them. Don’t trust the White House medical scientists.
I absolutely believe that all ideas and theories need to be questioned and debated, and the court of public opinion is one venue for airing these discussions. But the fact that all these pieces of research contained such fundamental flaws in logic and methodology damaged the cause of public health at this crucial moment in the pandemic. Together, these studies created an opening for an intellectually dishonest assertion, one that suggested that no one really knew anything about the virus, so anyone could be right and anyone could be wrong. What this perspective crucially left out is the fact that science operates on the principle of “Let’s test these ideas to see how right or wrong they are and under what circumstances these judgments are valid.”
Despite the failings in these studies, their sources possessed impressive enough pedigrees and carried enough reputational weight to cast the studies as definitive, giving them the power to muddy the waters just enough, to sow just the right amount of doubt. Perhaps because I had been outside the United States for so many years, working for PEPFAR, I was initially a half step behind in recognizing the harm that could be done by weak science such as this. In a post-truth America, it seemed, some in the White House, and pockets of Americans, would use any shred of evidence, even at this early stage, to undermine the public health case of the seriousness of Covid-19 to specific vulnerable groups, to make the objective somehow subjective.
I saw then (and I see now) why the public was confused by the conflicting messages coming out of the White House and the public health agencies. Some of the senior advisors in the West Wing had deliberately sowed and were quickly harvesting a crop of disinformation that outgrew and overshadowed the deliberative, careful data collection and analyses we had done. Just as these theories and justifications took root in the White House, so they did across the country, creating alternative interpretations of the science that would have profound implications for future mitigation efforts—from masking to testing to reducing indoor gatherings.
Once doubts over the data-driven, science-led response had crept in, the floodgates opened for exploiting ambiguities within the data and creating a parallel “data-driven” alternative reality. A pervasive attitude of “No one knows for sure, so do what you want” soon spread around the country.
Meanwhile, as these flawed but well-pedigreed studies arrived, other, subtler misrepresentations of the science and data had been pouring in from all sides. White House senior advisors repeatedly inundated me with published reports of various tenuously related pandemic topics from many sources. I believe many, like General Kellog, provided these to be helpful, to ensure I was seeing everything, but others used them to specifically undermine what I was saying and what I was asking them to do.
As the CEA report had shown, I was no longer the only one in the room armed with data to support her arguments. But not all data is created equal. This kind of blurring of the lines between complete data, warts and all, with its explanations for gaps and biases, and cherry-picked, incomplete data specially assembled to support a preconceived idea or theory, was the most dangerous area for a scientist to sink into.
An effective pandemic response has to build on a foundational bedrock of truth or, at a minimum, a shared understanding from which each side builds its argument. In the case of this pandemic, everyone had to agree that Covid-19 was a major risk to the health and well-being of the American public. Simply put, sizable influential elements in the Trump administration did not believe this about the virus in January, and they did not believe it about Covid-19 disease in April. This was in spite of what had happened in Italy and in spite of the devastation currently reaching a fever pitch in New York City and New Jersey, which we were all witnessing in real time. When we should have been debating the finer points of strategy, most of the Trump administration—and crucially, President Trump himself—were again arguing that the problem wasn’t greater than flu. The faulty studies and the memos from the CEA successfully gave my opposition more to work with.
The Trump administration hadn’t believed in the risk before I arrived. Over the course of March, I’d done everything I could to build consensus about the substantial risks to specific groups of Americans. I viewed getting two successive periods of shutdown as evidence that, on some level, I’d been able to break through with this message. But these new developments laid bare that any consensus that had been achieved was fleeting. Despite mounting evidence to the contrary, many in the Trump White House hadn’t changed their mind about the risks; it had just taken time for the opposing side to figure out how to counter our public health arguments and approach.
While there were undoubtedly people in the White House who viewed Covid-19 as a major risk, they were in the minority. I saw the entire NSC take the virus seriously (as did the vice president), and I believe to this day that Jared Kushner and his team saw the reality of the pandemic. The NSC had seen the early reports out of China and Asia before my arrival. Indeed, through Matt Pottinger, it was they who had recruited me to the White House to reinforce their warnings. While they didn’t play as active a role as Vice President Pence—he always listened to me and made every call to governors on my “need-to-call” list—to ensure they took the threat of the virus seriously. They also saw that solutions existed but that combating the misinformation and divergent points of view was as much a battle as trying to contain the pandemic. How can you combat something effectively if you can’t agree that it is actually a threat?
Despite these many frustrations, at least we were getting validation from others in the scientific community that our projections were accurate. At some point in the midst of all this back-and-forth, Tony passed along to me an email he’d randomly received from a highly credentialed statistician who’d run the numbers himself and come to the same conclusion we had. It was helpful intellectual support, reinforcing that we were correct, but it didn’t lessen the feeling of menace I sensed around me.
Tony and I connected nearly daily, week after week. I made sure he was seeing in the data what I was seeing. It was critical to me that we saw the same evidence the same way. This was true for Seema Verma, Bob Redfield, and Steve Hahn, too. The doctors’ group met three to four times a week, and I spoke with Bob and Steve day after day. I don’t think they ever wavered in grasping the seriousness of the pandemic or the need to do everything we could to battle the misinformation both within the White House and without.
In some of our discussions in the spring and carrying over into the early summer, Bob and I addressed a topic that was much discussed and debated from the first revelation of the outbreak in Wuhan. Because several research facilities located there were actively studying coronaviruses, much speculation went on about whether or not the wet market was definitively the point of origin of the outbreak. Bob and I would note that this virus, unlike other SARS strains, was unusually fit to adapting to a human host. Often with zoonotic viruses in that first jump from animals to humans it takes it awhile to adapt to its new host, but SARS-CoV-2 was unusually adapted to humans with high infectivity in the first surge unlike the prior SARS-CoV-1. Its unique characteristics presented us with real challenges but I wasn’t engaged in answering one of them—determining its origins. Once it was out and active, my priority was to save lives and though we had the genetic sequence (not the samples themselves, mind you) from the Chinese later than we would have liked, once we had it using that information took priority over any other consideration.
Both Bob and I were aware of other unintentional leaks or contamination issues arising out of labs around the world. While not common, they did occur despite safety measures being in place to prevent them. Whether that was the case here, we couldn’t say for sure, but it was possible. As I later told a house subcommittee investigating the full scope of the pandemic response, I didn’t have a definitive answer to their question regarding the origin of the virus. I did tell them that it was possible to get that answer, but it depended upon the Chinese releasing the very first samples of the virus taken from those who were first infected. It would be necessary to study the evolution of the virus in those first moments and study all of the original strains. Without having those samples of the original strains of the virus, it would be very difficult to ascertain with any degree of certainty whether it came from the wet market or a lab.
In either case, I don’t think that people were exposed to it intentionally. Entering that debate would be another distraction.
Steve Hahn and I bore the brunt of the intellectual assault that was hydroxychloroquine. Studies regarding its potential use as a Covid-19 therapeutic were passed along to both of us. These came out of France, China, and the United States, both from doctors and from people who had access to the internet but not a firm grasp of scientifically determining the drug’s efficacy as a treatment. Peter Navarro and television’s Dr. Mehmet Oz were among those pushing for us to back their belief that a drug used primarily for the treatment of malaria, and a sister drug, used to treat systemic lupus, rheumatoid arthritis, and other autoimmune conditions, should be used to treat Covid-19 disease.
In a crisis, taking proactive measures is important, and at the beginning I saw the FDA’s “emergency use authorization” (EUA) of hydroxychloroquine as falling into that category of proactive measures. So was continuing to definitively test to see if the drug would prove to be, in time, a therapeutic we should adopt and use more widely.
While the debate about the use of the drug sucked up too much of my time and attention, it did far more damage than that. Instead of the president’s delivering a consistent message about what we knew were effective mitigation measures and effective treatments, he was more preoccupied with touting the benefits of unproven, untested, potentially counterproductive drugs that had been brought to his attention daily by his inner circle. On March 21, the president tweeted about using hydroxychloroquine in combination with the antibiotic azithromycin, calling it a potential “game changer” and urging that the malaria drug be put to immediate use. A day later, he contradicted Tony, who had replied “The answer is no” to a journalist’s question about hydroxychloroquine’s effectiveness. Tony had gone on to state that the evidence for the drug’s being effective was “anecdotal.” The president said he felt good about it—that’s all it is, “just a feeling, you know” and “we’ll see what happens.”
Not everyone was able to sidestep hydroxychloroquine. Dr. Rick Bright, who, before being ousted, was in charge of vaccine development at the Biomedical Advanced Research and Development Authority (BARDA), a division of HHS, claimed that he had been dismissed for pushing back against what he called “misguided directives” to advocate for the use of hydroxychloroquine and chloroquine. He cited this among other examples of how the federal pandemic response was being interfered with.
Trump’s public cheerleading for hydroxychloroquine blurred important distinctions in some cases and caused direct harm in others. An Arizona man, having heard the president’s message on television, took a product that contained a form of hydroxychloroquine and died. The American Heart Association, the American College of Cardiology, and the Heart Rhythm Society warned that the combination of hydroxychloroquine and chloroquine might not be appropriate for patients with existing heart problems.
Later on, The Lancet issued a report on the efficacy of a variety of drugs that could be used to treat Covid-19 disease, including hydroxychloroquine. I was worried about this study’s validity due to the difficulties in global data collection I had seen. When I was asked about the Lancet piece, I said that it had some value. The study clearly clarified who was suffering and dying the most from SARS-CoV-2 infections: people with comorbidities. I thought my statement was clear, but the press later falsely reported that I supported the use of hydroxychloroquine. Why? Because I had said the article was of value—but in this, I had been referring to the demographic analysis showing that it was older patients and those with comorbidities who died from Covid-19 disease. I never said I thought the paper offered definitive support for or argued against the use of the drug or any other treatment. Such studies were under way in the United States and would produce definitive results down the line.
Within weeks of publication, the Lancet report was retracted. Additional reviews of the data showed that it was unreliable with regard to the interpretation of treatment. What wasn’t retracted was the claim that I supported the use of hydroxychloroquine. Understandably, this created more confusion. Did the White House and the Coronavirus Task Force believe in the effectiveness of the drug or not? The president and some of his advisors may have, but the rest of us were waiting for a definitive study we could trust.
The FDA approved the provisional use of hydroxychloroquine in hospitals and clinical trials only, stating again that the drug hadn’t been shown to be “safe and effective for treating or preventing COVID-19.” Despite this qualifier, twenty-two U.S. states stockpiled nearly thirty million doses. A May 11 study of more than fourteen hundred Covid-19 patients hospitalized in and around New York City found that those who took hydroxychloroquine with an antibiotic were twice as likely to experience cardiac arrest.
By mid-May, the president announced that, despite the concerns and warnings that hydroxychloroquine should be used only in hospital settings, he was taking the drug preventively and had been for ten days. Upon completing this two-week regimen following a viral exposure, he proclaimed not just that the drug hadn’t killed him, but that it had received “tremendous, rave reviews.” Though I was used to hearing hyperbole from the president, this felt particularly egregious. The reviews that mattered—what doctors and researchers were still working on—hadn’t come out yet.
By June, Steve Hahn and the FDA had reversed their position on “hydroxy” and revoked their EUA. It’s important to note the language used in their announcement: “Additionally, in light of ongoing serious cardiac adverse events and other potential serious side effects, the known and potential benefits of chloroquine and hydroxychloroquine no longer outweigh the known and potential risks for the authorized use.” By anyone’s standards, this was a reasonable and measured response.
Peter Navarro, one of President Trump’s trade advisors, who assisted in distributing the drugs, made even more plain what the administration thought of scientific rigor in commenting on this FDA rethink: “This is a Deep State blindside by bureaucrats who hate the administration they work for more than they want to save lives.”
What the FDA and Steve Hahn had done was exactly what one does in the scientific community: work to confirm the truth or falsity of a position. You examine the results. You make a decision based on the data.
What Navarro was actually saying was This administration hates all data that doesn’t support the unproven contentions we believe in that come from non-peer-reviewed anecdotal reports. Peter was knowledgeable and passionate about manufacturing and the need for the United States to reestablish the manufacturing of critical PPE and essential medication—an important point—but he also brought that same passion to the hydroxychloroquine debate. Peter believed so strongly in the data supporting hydroxychloroquine use, that he came to task force meetings armed with sheet after sheet of studies supporting his position. The first week in April, he got into a loud argument with Tony. Standing behind me and leaning over my shoulder, he angrily waved a stack of papers over my head toward Tony, shouting, “Here is the evidence this works! You’re ignoring all the data, and it’s killing people!”
There were other moments like this, when people passionately engaged with one another. Sometimes the disputes were based in evidence; other times, on peripheral unsupported positions. Some people in the White House seemed to believe that when reason failed, passion might carry the day. Peter Navarro crossed a line.
Steve, Tony, Bob, and I (and others) never took any of this personally; nor did any of us dislike any members of the administration. We hated the lies and distortions that were supported. We were deeply concerned when they used small, poorly designed investigations to arrive at definitive answers. We were enormously frustrated when “debates” took time away from the hard work of getting the data to avoid continually confusing the public.
Even with all the evidence in hand, the FDA continued its studies, concluding, on July 1, that “serious heart rhythm problems and other safety issues, including blood and lymph system disorders, kidney injuries, and liver problems and failure” were a consequence of using this drug. Tony and the NIH had done their own study and come to the same conclusion. Assistant Secretary for Health Brett Giroir also stated that he couldn’t recommend the drug’s use.
Meanwhile, President Trump’s conclusion was this: “Hydroxy has tremendous support, but politically it is toxic. If I would have said, ‘Do not use hydroxychloroquine under any circumstances,’ they would have come out and said it’s a great thing.”
Because Steve Hahn, as FDA commissioner, was the one directly involved, he did the talking for the rest of us in rebutting the president’s disdain. Speaking on ABC News, he said, “This is about science and data. There are randomized trials that show it doesn’t work.”
Real science and data are great things when they’re used transparently in open debate.